Evaluation method for fusion algorithm of autonomous landing guide information of shipboard aircraft

Abstract

The invention discloses an evaluation method for a shipboard aircraft autonomous landing guide information fusion algorithm. The evaluation method comprises the steps of 1, obtaining indexes of a plurality of fusion algorithms of shipboard aircraft autonomous landing guide information; step 2, establishing a comment set corresponding to indexes of each fusion algorithm; 3, obtaining a single-factor membership degree matrix of the shipboard aircraft autonomous landing guide information fusion algorithm by using a single-factor fuzzy evaluation method according to the weight distribution set and the comment set of the indexes of each fusion algorithm; 4, performing fuzzy synthesis on the single-factor membership degree matrix, and evaluating the advantages and disadvantages of each algorithm according to a score value; according to the method, the performance of the shipboard aircraft autonomous landing guide information fusion algorithm based on different technical principles can be effectively analyzed and evaluated, and the optimal scheme of the shipboard aircraft autonomous landing guide information fusion algorithm is selected.

Description

Technical Field

[0001] The present invention belongs to the field of evaluation of guidance information fusion algorithms, and in particular relates to an evaluation method for an autonomous landing guidance information fusion algorithm for carrier-based aircraft. Background Art

[0002] During the landing process, carrier-based aircraft will inevitably face severe external interference such as atmospheric turbulence, deck wind, and ship wake. In addition, factors such as narrow landing space and floating decks greatly increase the difficulty and risk of operation for pilots who have experienced long-term combat flights, whether in terms of physical fitness, attention or mental state. As for drones, their role on the battlefield will become increasingly important in future wars, and autonomous landing guidance technology, as a prerequisite for drones to board ships, has received urgent demand and widespread attention.

[0003] Aircraft guided by different single sensors are not applicable due to different factors such as weather, distance, and signal transmission environment, which results in impaired guidance accuracy or excessive error when guidance instruction information is subject to strong interference signals. Therefore, the redundant guidance information fusion algorithm is used to calibrate, combine, correlate, and synthesize the guidance information of multiple guidance methods, and remove wild values, as well as handle the special situation of a single sensor failure. The guidance information of different sensors is dynamically weighted, which plays a vital role in achieving stable and reliable autonomous landing guidance. There are many technical solutions to achieve redundant guidance information fusion. Due to different technical implementation principles, different technical solutions have different applicability and inapplicability. In order to improve the reliability and stability of the redundant guidance information fusion algorithm for autonomous landing, a comparative test is conducted by combining simulation and experiment, and the most suitable technical solution is selected scientifically, accurately, and comprehensively to achieve the purpose of autonomous landing guidance.

[0004] The existing technology lacks a standard and complete comparative test index system. Therefore, a systematic and systematic redundancy-guided information fusion algorithm comparative test technology is proposed to achieve the selection of the optimal algorithm solution. Summary of the invention

[0005] The purpose of the present invention is to provide an evaluation method for the information fusion algorithm of autonomous landing guidance of carrier-based aircraft, which can scientifically, accurately and comprehensively select the most suitable technical solution to achieve the purpose of autonomous landing guidance.

[0006] The present invention adopts the following technical solution: a method for evaluating an information fusion algorithm for autonomous landing guidance of a carrier-based aircraft, comprising:

[0007] Step 1: Obtain indicators of multiple fusion algorithms for carrier-based aircraft autonomous landing guidance information;

[0008] Step 2: Establish a comment set corresponding to the indicators of each fusion algorithm;

[0009] Step 3: According to the weight distribution set and comment set of the indicators of each fusion algorithm, the single factor fuzzy evaluation method is used to obtain the single factor membership matrix of the carrier-based aircraft autonomous landing guidance information fusion algorithm;

[0010] Step 4: Perform fuzzy synthesis on the single-factor membership matrix and evaluate the pros and cons of each algorithm based on the score value.

[0011] Furthermore, the indicators in step 1 include: real-time performance, clutter quantity, robustness, fault tolerance, accuracy, packet loss accuracy and average error of landing point.

[0012] Furthermore, the comment set in step 2 is:

[0013]

[0014]

[0015] Furthermore, the weight distribution set in step 3 is:

[0016]

[0017] Furthermore, the comment V1 in step 2 adopts a small-sized distribution, namely:

[0018]

[0019] Wherein, a is the lower bound of the quantization interval of V1 comments, b is the upper bound of the quantization interval of V1 comments; x is the test value of the jth test indicator of the i-th test content;

[0020] The V2, V3, and V4 comments in step 2 adopt an intermediate distribution, namely:

[0021]

[0022] Among them, a1 is the lower bound of the quantization interval of V2 comments, b1 is the upper bound of the quantization interval of V2 comments, c is the upper bound of the quantization interval of V3 comments, and d is the upper bound of the quantization interval of V4 comments; x is the test value of the jth test indicator of the i-th test content;

[0023] In step 2, a large-scale distribution is used for V5 comments, namely:

[0024]

[0025] Among them, a2 is the lower limit of the comment V5 quantization interval, b2 is the upper limit of the V5 comment quantization interval, and x is the test value of the jth test indicator of the i-th test content.

[0026] The beneficial effects of the present invention are:

[0027] The present invention establishes a complete indicator system including real-time performance, clutter quantity, robustness, fault tolerance, accuracy, anti-interference, packet loss accuracy, and average landing point error, clarifies the meaning and technical method of each indicator, and combines the fuzzy comprehensive evaluation method to perform weight allocation, which can effectively analyze and evaluate the performance of carrier-based aircraft autonomous landing guidance information fusion algorithms based on different technical principles, and realize the selection of the optimal solution for the carrier-based aircraft autonomous landing guidance information fusion algorithm. DETAILED DESCRIPTION

[0028] The present invention is described in detail below in conjunction with specific implementation modes.

[0029] The present invention discloses a method for evaluating an autonomous carrier-based aircraft landing guidance information fusion algorithm, comprising:

[0030] Step 1: Obtain indicators of multiple fusion algorithms for carrier-based aircraft autonomous landing guidance information

[0031] First, determine the factor set of the evaluation object

[0032] The factor set of the evaluation object is established as U,

[0033] U={u1,u2,…,u7}

[0034] where u i is the i-th sub-goal of the test content layer in the indicator system, u i = {u i1 ,…,u ip}, where i = 1, 2, ..., 7, and p is the number of indicators in the test indicator layer in the indicator system.

[0035] The test index is calculated as follows:

[0036] 1. Real-time performance: Analyze the phase difference between the excitation signal and the fused output signal. The time delay can be calculated by combining the phase difference with the excitation signal frequency.

[0037]

[0038] in, is the excitation signal phase, is the phase of the fused output signal.

[0039] 2. The number of clutter: In the process of redundant guidance information fusion guidance, the fusion algorithm needs to fuse different types of sensors with the same timestamp. During data transmission, there will always be deviations in the arrival time of data. Different fusion algorithms may cause different degrees of clutter. Clutter has a frequency different from the main frequency. If the frequency of the clutter is within the response frequency of the control system, it is easy to affect the flight control system. Therefore, the number of clutter is an indicator for evaluating algorithm performance.

[0040] η=k

[0041] Among them, η is the number of clutter and k is the number of different frequencies.

[0042] 3. Robustness: During the autonomous landing process, the ship is in motion, especially near the stern area, and the aircraft needs to be guided to follow the movement of the ship. The glide angle will change. Therefore, it is necessary to consider the change of the glide angle and whether the fusion algorithm is robust.

[0043] The guidance information of the redundant guidance information fusion algorithm, the longitudinal deviation robustness is solved by the following formula:

[0044]

[0045] The lateral deviation robustness is solved by the following equation:

[0046]

[0047] The overall robustness of the guided information can be expressed as:

[0048] rob=|σ Rob_Y -σ y |+|σ Rob_H -σ h |

[0049] Among them, N0 is the number of repeated experiments, M0 is the sample size of a single experiment, is the longitudinal deviation of the fused output signal, y i is the longitudinal deviation of the true value, is the lateral deviation of the fused output signal, h i is the lateral deviation from the true value.

[0050] 4. Fault tolerance: During the fusion process, there may be sudden sensor disconnection or failure, so it is necessary to judge the fault tolerance of sensor data. The types of sensors that generate guidance information sources include satellite navigation sensors, radar sensors, and photoelectric sensors. By maintaining the normal guidance of inertial navigation, the single fault of each sensor and the faults of each sensor are tested separately, the output results of the fusion algorithm are analyzed, and the difference between the input result and the output result is used for LBQ test, and the fault tolerance capability is identified.

[0051] The LBQ calculation index can be expressed as:

[0052]

[0053] Where n is the number of samples, is the correlation coefficient of the sample k-order lag, and this statistic follows a chi-square distribution with h degrees of freedom.

[0054] 5. Accuracy: Ideally, for the fusion algorithm, when the short transition process ends, stability should be achieved and the system state estimate should be consistent with the standard value. However, in reality, due to inaccurate system modeling, there will be errors between the system state estimate and the standard value. Therefore, the accuracy index must be established in the algorithm quality evaluation.

[0055] The lateral accuracy can be expressed as:

[0056]

[0057] The longitudinal accuracy can be expressed as:

[0058]

[0059] In the formula, N is the number of repeated tests, M is the sample size of a single test, is the longitudinal deviation of the fused output signal, y i is the longitudinal deviation of the true value, is the lateral deviation of the fused output signal, h i is the lateral deviation from the true value.

[0060] 6. Packet loss accuracy: Packet loss is inevitable when the guidance information is transmitted through the data link. Therefore, it is necessary to test and analyze this capability to determine the impact of packet loss on the fusion guidance accuracy during data transmission.

[0061] The lateral accuracy under the condition of launching 2 and losing 1 can be calculated by the following formula:

[0062]

[0063] In the formula, It is the lateral deviation output by the fusion algorithm under the condition of 2 transmission and 1 loss;

[0064] The longitudinal accuracy under the condition of 2 launches and 1 loss can be calculated by the following formula:

[0065]

[0066] In the formula, It is the longitudinal deviation output by the fusion algorithm under the condition of 2 transmission and 1 loss;

[0067] The comprehensive accuracy under the condition of 2 launches and 1 loss can be calculated by the following formula:

[0068] drop2=|σ y_drop2 -σ y |+|σ h_drop2 -σ h |

[0069] The lateral accuracy under the condition of 4 launches and 1 loss can be calculated by the following formula:

[0070]

[0071] In the formula, It is the lateral deviation output by the fusion algorithm under the condition of 4 transmission and 1 loss;

[0072] The longitudinal accuracy under the condition of 4 launches and 1 loss can be calculated by the following formula:

[0073]

[0074] In the formula, It is the longitudinal deviation output by the fusion algorithm under the condition of 2 transmission and 1 loss;

[0075] The comprehensive accuracy under the condition of 4 launches and 1 loss can be calculated by the following formula:

[0076] drop4=|σ y_drop4 -σ y |+|σ h_drop4 -σ h |.

[0077] 7. Average landing point error: After the autonomous landing process is completed, the landing point will inevitably deviate from the ideal landing point, and the reasons for this deviation are relatively wide, including the original accuracy of the sensor, the stability and reliability of the fusion algorithm, the control performance of the flight control system, etc. If the deviation is too large and exceeds the safety range, it will affect the safety of the landing and the success of the autonomous landing mission. It is necessary to test whether the fusion algorithm can effectively reduce the deviation between the landing point and the ideal landing point.

[0078] The X-direction error of the landing point is:

[0079]

[0080] The Y direction error of the landing point is:

[0081]

[0082] In the formula, is the error of the landing point in the X direction output by the fusion algorithm; end is the error of the actual landing point of the carrier-based aircraft in the X direction; is the error of the landing point in the Y direction output by the fusion algorithm; endis the error of the actual landing point of the carrier-based aircraft in the Y direction.

[0083] Step 2: Establish a comment set V corresponding to the indicators of each fusion algorithm, V = {V1, V2, V3, V4, V5}.

[0084] The quantization intervals of the comment set are shown in the following table:

[0085]

[0086]

[0087] Step 3: According to the weight distribution set and comment set of the indicators of each fusion algorithm, the single factor fuzzy evaluation method is used to obtain the single factor membership matrix of the carrier-based aircraft autonomous landing guidance information fusion algorithm.

[0088] The Analytic Hierarchy Process (AHP) is used to determine the weight distribution set W of the fusion algorithm performance sub-goals and the weight distribution set W of the evaluation index. i .

[0089] W={W1,W2,…,W7}

[0090] W i ={w i1 ,…,w ip}

[0091] Among them, W i and w ij Satisfy: 0≤W i ,w ij ≤1,

[0092] Where, {W1, W2, …, W7} are the weight distributions of guiding the real-time performance (u1), the number of clutter (u2), robustness (u3), fault tolerance (u4), accuracy (u5), packet loss accuracy (u6), and the average error of the landing point (u7) of the fusion algorithm; {w i3 ,w i2 ,…,w ip} are the weight allocations of the p indicators in the next layer of {W1,W2,…,W7} respectively.

[0093] In terms of weight determination, since there are many assessment indicators, approximate equal weights are adopted, and the fusion algorithms are compared and analyzed in combination with the fuzzy comprehensive evaluation method to reduce subjective factors as much as possible.

[0094] Table 1 Weight distribution set of indicators of each fusion algorithm

[0095]

[0096] Construct a single-factor membership matrix:

[0097]

[0098] Among them, the matrix element r ijk Indicates the degree to which the jth test indicator of the i-th test content belongs to the k-th comment set.

[0099] For comment V1, a small distribution is used, namely:

[0100]

[0101] Among them, a is the lower limit of the quantization interval of V1 comments, b is the upper limit of the quantization interval of V1 comments; x is the test value of the jth test indicator of the i-th test content.

[0102] For V2, V3, and V4 comments, an intermediate distribution is used, namely:

[0103]

[0104] Among them, a1 is the lower limit of the V2 comment quantization interval, b1 is the upper limit of the V2 comment quantization interval, c is the upper limit of the V3 comment quantization interval, and d is the upper limit of the V4 comment quantization interval; x is the test value of the jth test indicator of the i-th test content.

[0105] For V5 comments, a relatively large distribution is used, namely:

[0106]

[0107] Among them, a2 is the lower limit of the comment V5 quantization interval, b2 is the upper limit of the V5 comment quantization interval, and x is the test value of the jth test indicator of the i-th test content.

[0108] Step 4: Perform fuzzy synthesis on the single-factor membership matrix and evaluate the pros and cons of each algorithm based on the score value.

[0109] First, use The operator performs fuzzy synthesis on the i-th test index:

[0110]

[0111] Then perform fuzzy synthesis on multiple indicators:

[0112]

[0113] To analyze the fuzzy comprehensive evaluation results, first normalize the multi-index fuzzy synthesis results:

[0114]

[0115] Secondly, assign values ​​to the comment set:

[0116] Evaluation Rating Table <![CDATA[Very good (V1)]]> <![CDATA[Better (V2)]]> <![CDATA[Medium (V3)]]> <![CDATA[Poor (V4)]]> <![CDATA[Very poor (V5)]]> Assignment 100 90 80 70 60

[0117] By multiplying the value of the comment set by the corresponding multi-index fuzzy synthesis result, the score of the fusion algorithm can be obtained and the quality of the fusion algorithm can be evaluated.

[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for evaluating an information fusion algorithm for autonomous carrier-based aircraft landing guidance, characterized in that: include: Step 1: Obtain indicators of multiple fusion algorithms for carrier-based aircraft autonomous landing guidance information; Step 2: Establish a comment set corresponding to the indicators of each fusion algorithm; Step 3: According to the weight distribution set and comment set of the indicators of each fusion algorithm, the single factor fuzzy evaluation method is used to obtain the single factor membership matrix of the carrier-based aircraft autonomous landing guidance information fusion algorithm; Step 4: Perform fuzzy synthesis on the single-factor membership matrix and evaluate the pros and cons of each algorithm based on the score value.

2. The evaluation method for the carrier-based aircraft autonomous landing guidance information fusion algorithm according to claim 1, characterized in that: The indicators in step 1 include: real-time performance, clutter quantity, robustness, fault tolerance, accuracy, packet loss accuracy, and average error of landing point.

3. The evaluation method for the carrier-based aircraft autonomous landing guidance information fusion algorithm according to claim 2, characterized in that: The set of comments in step 2 is:

4. The evaluation method for the carrier-based aircraft autonomous landing guidance information fusion algorithm according to claim 2, characterized in that: The weight distribution set in step 3 is:

5. The evaluation method for the carrier-based aircraft autonomous landing guidance information fusion algorithm according to claim 3, characterized in that: The comment V1 in step 2 adopts a small distribution, that is: Wherein, a is the lower bound of the quantization interval of V1 comments, b is the upper bound of the quantization interval of V1 comments; x is the test value of the jth test indicator of the i-th test content; The V2, V3, and V4 comments in step 2 adopt an intermediate distribution, namely: Among them, a1 is the lower bound of the quantization interval of V2 comments, b1 is the upper bound of the quantization interval of V2 comments, c is the upper bound of the quantization interval of V3 comments, and d is the upper bound of the quantization interval of V4 comments; x is the test value of the jth test indicator of the i-th test content; In step 2, a large-scale distribution is used for V5 comments, namely: Among them, a2 is the lower limit of the comment V5 quantization interval, b2 is the upper limit of the V5 comment quantization interval, and x is the test value of the jth test indicator of the i-th test content.

Images